arXiv subjects

Xiaowei Zhao

Publications and source records attributed to Xiaowei Zhao.

At least 37 records · Page 2Linked to original sources

arXiv2023

Wake effects of offshore wind farm clusters revealed by SAR and WRF

Wake effects, i.e. the reduced momentum and increased turbulence caused by the upstream wind farm, have a significant adverse impact on downstream wind farms. However, due to the lack of ground truth for flow scenarios without wind farms in place (as the wind farm has already been constructed on site), it is extremely difficult to quantify the real impact caused by the presence of upstream wind farms for the downstream area. This paper seeks to develop a potential solution by taking advantage of both SAR and WRF. Specifically, the real-world wind speed with wind farms is retrieved from the SAR images using the C-band model, while the scenario without wind farms is simulated by the WRF model. By combining these two technologies, the potential impact of long-distance wind farm wakes is revealed and analysed.

physics.ao-ph↗

arXiv2023

Observations of Mini Coronal Dimmings Caused by Small-scale Eruptions in the Quiet Sun

Small-scale eruptions could play an important role in coronal heating, generation of solar energetic particles (SEPs), and mass source of the solar wind. However, they are poorly observed, and their characteristics, distributions, and origins remain unclear. Here a mini coronal dimming was captured by the recently launched Solar Orbiter spacecraft. The observations indicate that a minifilament eruption results in the dimming and takes away approximately $(1.65\pm0.54)\times10^{13}$ g of mass, which also exhibits similar features as the sources of SEP events. The released magnetic free energy is of the order of $\sim10^{27}$ erg. Our results suggest that weak constraining force makes the flux rope associated with the minifilament easily enter a torus-unstable domain. We discuss that weak magnetic constraints from low-altitude background fields may be a general condition for the quiet-Sun eruptions, which provide a possible mechanism for the transport of coronal material and energy from the lower to the middle or even higher corona.

astro-ph.SR↗

arXiv2023

LSwinSR: UAV Imagery Super-Resolution based on Linear Swin Transformer

Super-resolution, which aims to reconstruct high-resolution images from low-resolution images, has drawn considerable attention and has been intensively studied in computer vision and remote sensing communities. The super-resolution technology is especially beneficial for Unmanned Aerial Vehicles (UAV), as the amount and resolution of images captured by UAV are highly limited by physical constraints such as flight altitude and load capacity. In the wake of the successful application of deep learning methods in the super-resolution task, in recent years, a series of super-resolution algorithms have been developed. In this paper, for the super-resolution of UAV images, a novel network based on the state-of-the-art Swin Transformer is proposed with better efficiency and competitive accuracy. Meanwhile, as one of the essential applications of the UAV is land cover and land use monitoring, simple image quality assessments such as the Peak-Signal-to-Noise Ratio (PSNR) and the Structural Similarity Index Measure (SSIM) are not enough to comprehensively measure the performance of an algorithm. Therefore, we further investigate the effectiveness of super-resolution methods using the accuracy of semantic segmentation. The code will be available at https://github.com/lironui/LSwinSR.

eess.IV↗

arXiv2022

GMF: General Multimodal Fusion Framework for Correspondence Outlier Rejection

Rejecting correspondence outliers enables to boost the correspondence quality, which is a critical step in achieving high point cloud registration accuracy. The current state-of-the-art correspondence outlier rejection methods only utilize the structure features of the correspondences. However, texture information is critical to reject the correspondence outliers in our human vision system. In this paper, we propose General Multimodal Fusion (GMF) to learn to reject the correspondence outliers by leveraging both the structure and texture information. Specifically, two cross-attention-based fusion layers are proposed to fuse the texture information from paired images and structure information from point correspondences. Moreover, we propose a convolutional position encoding layer to enhance the difference between Tokens and enable the encoding feature pay attention to neighbor information. Our position encoding layer will make the cross-attention operation integrate both local and global information. Experiments on multiple datasets(3DMatch, 3DLoMatch, KITTI) and recent state-of-the-art models (3DRegNet, DGR, PointDSC) prove that our GMF achieves wide generalization ability and consistently improves the point cloud registration accuracy. Furthermore, several ablation studies demonstrate the robustness of the proposed GMF on different loss functions, lighting conditions and noises.The code is available at https://github.com/XiaoshuiHuang/GMF.

cs.CV↗

arXiv2022

Revisiting Open World Object Detection

Open World Object Detection (OWOD), simulating the real dynamic world where knowledge grows continuously, attempts to detect both known and unknown classes and incrementally learn the identified unknown ones. We find that although the only previous OWOD work constructively puts forward to the OWOD definition, the experimental settings are unreasonable with the illogical benchmark, confusing metric calculation, and inappropriate method. In this paper, we rethink the OWOD experimental setting and propose five fundamental benchmark principles to guide the OWOD benchmark construction. Moreover, we design two fair evaluation protocols specific to the OWOD problem, filling the void of evaluating from the perspective of unknown classes. Furthermore, we introduce a novel and effective OWOD framework containing an auxiliary Proposal ADvisor (PAD) and a Class-specific Expelling Classifier (CEC). The non-parametric PAD could assist the RPN in identifying accurate unknown proposals without supervision, while CEC calibrates the over-confident activation boundary and filters out confusing predictions through a class-specific expelling function. Comprehensive experiments conducted on our fair benchmark demonstrate that our method outperforms other state-of-the-art object detection approaches in terms of both existing and our new metrics. Our benchmark and code are available at https://github.com/RE-OWOD/RE-OWOD.

cs.CV↗

arXiv2021

IMFNet: Interpretable Multimodal Fusion for Point Cloud Registration

The existing state-of-the-art point descriptor relies on structure information only, which omit the texture information. However, texture information is crucial for our humans to distinguish a scene part. Moreover, the current learning-based point descriptors are all black boxes which are unclear how the original points contribute to the final descriptor. In this paper, we propose a new multimodal fusion method to generate a point cloud registration descriptor by considering both structure and texture information. Specifically, a novel attention-fusion module is designed to extract the weighted texture information for the descriptor extraction. In addition, we propose an interpretable module to explain the original points in contributing to the final descriptor. We use the descriptor element as the loss to backpropagate to the target layer and consider the gradient as the significance of this point to the final descriptor. This paper moves one step further to explainable deep learning in the registration task. Comprehensive experiments on 3DMatch, 3DLoMatch and KITTI demonstrate that the multimodal fusion descriptor achieves state-of-the-art accuracy and improve the descriptor's distinctiveness. We also demonstrate that our interpretable module in explaining the registration descriptor extraction.

cs.CV↗

arXiv2020

Simplifying Reinforced Feature Selection via Restructured Choice Strategy of Single Agent

Feature selection aims to select a subset of features to optimize the performances of downstream predictive tasks. Recently, multi-agent reinforced feature selection (MARFS) has been introduced to automate feature selection, by creating agents for each feature to select or deselect corresponding features. Although MARFS enjoys the automation of the selection process, MARFS suffers from not just the data complexity in terms of contents and dimensionality, but also the exponentially-increasing computational costs with regard to the number of agents. The raised concern leads to a new research question: Can we simplify the selection process of agents under reinforcement learning context so as to improve the efficiency and costs of feature selection? To address the question, we develop a single-agent reinforced feature selection approach integrated with restructured choice strategy. Specifically, the restructured choice strategy includes: 1) we exploit only one single agent to handle the selection task of multiple features, instead of using multiple agents. 2) we develop a scanning method to empower the single agent to make multiple selection/deselection decisions in each round of scanning. 3) we exploit the relevance to predictive labels of features to prioritize the scanning orders of the agent for multiple features. 4) we propose a convolutional auto-encoder algorithm, integrated with the encoded index information of features, to improve state representation. 5) we design a reward scheme that take into account both prediction accuracy and feature redundancy to facilitate the exploration process. Finally, we present extensive experimental results to demonstrate the efficiency and effectiveness of the proposed method.

cs.LG↗

arXiv2020

Characteristics and Importance of "ICME-in-Sheath" Phenomenon and Upper Limit for Geomagnetic Storm Activity

As an important source for large geomagnetic storms, an "ICME-in-sheath" is a completely shocked interplanetary coronal mass ejection (ICME) stuck in the sheath between a shock and host ejecta. Typical characteristics are identified from coordinated multi-sets of observations: (1) it is usually short in duration and lasts a few hours at 1 AU; (2) its solar wind parameters, in particular the magnetic field, seem to keep enhanced for a large range of distances; and (3) common ICME signatures are often lost. The host ejecta could be a single ICME or a complex ejecta, being fast enough to drive a shock. These results clarify previous misinterpretations of this phenomenon as a normal part of a sheath region. The "ICME-in-sheath" phenomenon, together with a preconditioning effect, produced an extreme set of the magnetic field, speed and density near 1 AU in the 2012 July 23 case, all around their upper limits at the same time. This is probably the most extreme solar wind driving at 1 AU and enables us to estimate the plausible upper limit for geomagnetic storm activity. With an appropriate modification in the southward field, we suggest that a geomagnetic storm with a minimum $D_{\rm st}$ of about $-2000$ nT could occur in principle. The magnetopause would be compressed to about 3.3 Earth radii from the Earth's center, well inside the geosynchronous orbit.

astro-ph.SR↗

arXiv2019

Characteristics of a Gradual Filament Eruption and Subsequent CME Propagation in Relation to a Strong Geomagnetic Storm

An unexpected strong geomagnetic storm occurred on 2018 August 26, which was caused by a slow coronal mass ejection (CME) from a gradual eruption of a large quiet-region filament. We investigate the eruption and propagation characteristics of this CME in relation to the strong geomagnetic storm with remote sensing and in situ observations. Coronal magnetic fields around the filament are extrapolated and compared with EUV observations. We determine the propagation direction and tilt angle of the CME flux rope near the Sun using a graduated cylindrical shell (GCS) model and the Sun-to-Earth kinematics of the CME with wide-angle imaging observations from STEREO A. We reconstruct the flux-rope structure using a Grad-Shafranov technique based on the in situ measurements at the Earth and compare it with those from solar observations and the GCS results. Our conclusions are as follows: (1) the eruption of the filament was unusually slow and occurred in the regions with relatively low critical heights of the coronal field decay index; (2) the axis of the CME flux rope rotated in the corona as well as in interplanetary space, which tended to be aligned with the local heliospheric current sheet; (3) the CME was bracketed between slow and fast solar winds, which enhanced the magnetic field inside the CME at 1 AU; (4) the geomagnetic storm was caused by the enhanced magnetic field and a southward orientation of the flux rope at 1 AU from the rotation of the flux rope.

astro-ph.SR↗

arXiv2019

Quantifying the Propagation of Fast Coronal Mass Ejections from the Sun to Interplanetary Space Combining Remote Sensing and Multi-Point in-situ Observations

In order to have a comprehensive view of the propagation and evolution of coronal mass ejections (CMEs) from the Sun to deep interplanetary space beyond 1 au, we carry out a kinematic analysis of 7 CMEs in solar cycle 23. The events are required to have coordinated coronagraph observations, interplanetary type II radio bursts, and multi-point in-situ measurements at the Earth and Ulysses. A graduated cylindrical shell model, an analytical model without free parameters and a magnetohydrodynamic model are used to derive CME kinematics near the Sun, to quantify the CME/shock propagation in the Sun-Earth space, and to connect in-situ signatures at the Earth and Ulysses, respectively. We find that each of the 7 CME-driven shocks experienced a major deceleration before reaching 1 au and thereafter propagated with a gradual deceleration from the Earth to larger distances. The resulting CME/shock propagation profile for each case is roughly consistent with all the data, which verifies the usefulness of the simple analytical model for CME/shock propagation in the heliosphere. The statistical analysis of CME kinematics indicates a tendency that the faster the CME, the larger the deceleration, and the shorter the deceleration time period within 1 au. For several of these events, the associated geomagnetic storms were mainly caused by the southward magnetic fields in the sheath region. In particular, the interaction between a CME-driven shock and a preceding ejecta significantly enhanced the preexisting southward magnetic fields and gave rise to a severe complex geomagnetic storm.

astro-ph.SR↗

arXiv2019

A Comparative Study of 2017 July and 2012 July Complex Eruptions: Are Solar Superstorms "Perfect Storms" in Nature?

It is paramount from both scientific and societal perspectives to understand the generation of extreme space weather. We discuss the formation of solar superstorms based on a comparative study of the 2012 July 23 and 2017 July 23 eruptions. The first one is Carrington-class, and the second could rival the 1989 March event that caused the most intense geomagnetic storm of the space age. Observations of these events in the historically weak solar cycle 24 indicate that a solar superstorm can occur in any solar cycle and at any phase of the cycle. Recurrent patterns are identified in both cases, including the long-lived eruptive nature of the active region, a complex event composed of successive eruptions from the same active region, and in-transit interaction between the successive eruptions resulting in exceptionally strong ejecta magnetic fields at 1 AU. Each case also shows unique characteristics. Preconditioning of the upstream solar wind leading to unusually high solar wind speeds at 1 AU is observed in the first case whereas not in the latter. This may suggest that the concept of "preconditioning" appears to be necessary for making a Carrington-class storm. We find a considerable deflection by nearby coronal holes in the second case but not in the first. On the basis of these results, we propose a hypothesis for further investigation that superstorms are "perfect storms" in nature, i.e., a combination of circumstances that results in an event of unusual magnitude. Historical records of some extreme events seem to support our hypothesis.

astro-ph.SR↗

arXiv2018

Geometry, Kinematics and Heliospheric Impact of a Large CME-driven Shock in 2017 September

A powerful coronal mass ejection (CME) occurred on 2017 September 10 near the end of the declining phase of the historically weak solar cycle 24. We obtain new insights concerning the geometry and kinematics of CME-driven shocks in relation to their heliospheric impacts from the optimal, multi-spacecraft observations of the eruption. The shock, which together with the CME driver can be tracked from the early stage to the outer corona, shows a large oblate structure produced by the vast expansion of the ejecta. The expansion speeds of the shock along the radial and lateral directions are much larger than the translational speed of the shock center, all of which increase during the flare rise phase, peak slightly after the flare maximum and then decrease. The near simultaneous arrival of the CME-driven shock at the Earth and Mars, which are separated by 156.6$^{\circ}$ in longitude, is consistent with the dominance of expansion over translation observed near the Sun. The shock decayed and failed to reach STEREO A around the backward direction. Comparison between ENLIL MHD simulations and the multi-point in situ measurements indicates that the shock expansion near the Sun is crucial for determining the arrival or non-arrival and space weather impact at certain heliospheric locations. The large shock geometry and kinematics have to be taken into account and properly treated for accurate predictions of the arrival time and space weather impact of CMEs.

astro-ph.SR↗

arXiv2018

A Solar Eruption with Relatively Strong Geo-effectiveness Originating from Active Region Peripheral Diffusive Polarities

We report the observations of a moderate but relatively intense geo-effective solar eruption on 2015 November 4 from the peripheral diffusive polarities of active region 12443. We use space-borne Solar Dynamics Observatory and ACE observations. EUV images identified helical pattern along a filament channel and we regard this channel as flux-rope structure. Flow velocity derived from tracked magnetograms infers converging motion along the polarity inversion line beneath the filament channel. An associated magnetic cancellation process was detected in the converging region. Further, the pre-eruptive EUV brightening was observed in the converging region, the most intense part of which appeared in the magnetic cancellation region. These observations imply that the converging and cancelling flux probably contributed to the formation of the helical magnetic fields associated with the flux rope. A filament-height estimation method suggests that the middle part of the filament probably lies at a low altitude and was consistent with the initial place of the eruption. A thick current channel associated with the flux rope is also determined. For a expanding thick current channel, the critical height of the decay index for torus instability lies in the range of 37 - 47 Mm. Southward magnetic fields in the sheath and the ejecta induced a geomagnetic storm with a Dst global minimum of ~-90 nT.

astro-ph.SR↗

arXiv2018

A Stealth CME Bracketed between Slow and Fast Wind Producing Unexpected Geo-effectiveness

We investigate how a weak coronal mass ejection (CME) launched on 2016 October 8 without obvious signatures in the low corona produced a relatively intense geomagnetic storm. Remote sensing observations from SDO, STEREO and SOHO and in situ measurements from WIND are employed to track the CME from the Sun to the Earth. Using a graduated cylindrical shell (GCS) model, we estimate the propagation direction and the morphology of the CME near the Sun. CME kinematics are determined from the wide-angle imaging observations of STEREO A and are used to predict the CME arrival time and speed at the Earth. We compare ENLIL MHD simulation results with in situ measurements to illustrate the background solar wind where the CME was propagating. We also apply a Grad--Shafranov technique to reconstruct the flux rope structure from in situ measurements in order to understand the geo-effectiveness associated with the CME magnetic field structure. Key results are obtained concerning how a weak CME can generate a relatively intense geomagnetic storm: (1) there were coronal holes at low latitudes, which could produce high speed streams (HSSs) to interact with the CME in interplanetary space; (2) the CME was bracketed between a slow wind ahead and a HSS behind, which enhanced the southward magnetic field inside the CME and gave rise to the unexpected geomagnetic storm.

physics.space-ph↗

arXiv2018

Multi-Spacecraft Observations of the Rotation and Non-Radial Motion of a CME Flux Rope causing an intense geomagnetic storm

We present an investigation of the rotation and non-radial motion of a coronal mass ejection (CME) from AR 12468 on 2015 December 16 using observations from SDO, SOHO, STEREO A and Wind. The EUV and HMI observations of the source region show that the associated magnetic flux rope (MFR) axis pointed to the east before the eruption. We use a nonlinear fore-free field (NLFFF) extrapolation to determine the configuration of the coronal magnetic field and calculate the magnetic energy density distributions at different heights. The distribution of the magnetic energy density shows a strong gradient toward the northeast. The propagation direction of the CME from a Graduated Cylindrical Shell (GCS) modeling deviates from the radial direction of the source region by about 45 degr in longitude and about 30 degr in latitude, which is consistent with the gradient of the magnetic energy distribution around the AR.The MFR axis determined by the GCS modeling points southward, which has rotated counterclockwise by about 95 degr compared with the orientation of the MFR in the low corona.The MFR reconstructed by a Grad-Shafranov (GS) method at 1 AU has almost the same orientation as the MFR from the GCS modeling, which indicates that the MFR rotation occurred in the low corona. It is the rotation of the MFR that caused the intense geomagnetic storm with the minimum Dst of -155 nT. These results suggest that the coronal magnetic field surrounding the MFR plays a crucial role in the MFR rotation and propagation direction.

astro-ph.SR↗

arXiv2017

Multi-modal Aggregation for Video Classification

In this paper, we present a solution to Large-Scale Video Classification Challenge (LSVC2017) [1] that ranked the 1st place. We focused on a variety of modalities that cover visual, motion and audio. Also, we visualized the aggregation process to better understand how each modality takes effect. Among the extracted modalities, we found Temporal-Spatial features calculated by 3D convolution quite promising that greatly improved the performance. We attained the official metric mAP 0.8741 on the testing set with the ensemble model.

cs.CV↗

arXiv2017

Propagation and Interaction Properties of Successive Coronal Mass Ejections in Relation to a Complex Type II Radio Burst

We examine the propagation and interaction properties of three successive coronal mass ejections (CMEs) from 2001 November 21-22, with a focus on their connection with the behaviors of the associated long-duration complex type II radio burst. In combination with coronagraph and multi-point in situ observations, the long-duration type II burst provides key features for resolving the propagation and interaction complexities of the three CMEs. The two CMEs from November 22 interacted first and then overtook the November 21 CME at a distance of about 0.85 AU from the Sun. The time scale for the shock originally driven by the last CME to propagate through the preceding two CMEs is estimated to be about 14 and 6 hr, respectively. We present a simple analytical model without any free parameters to characterize the whole Sun-to-Earth propagation of the shock, which shows a remarkable consistency with all the available data and MHD simulations even out to the distance of Ulysses (2.34 AU). The coordination of in situ measurements at the Earth and Ulysses, which were separated by about $71.4^{\circ}$ in latitude, gives important clues for the understanding of shock structure and the interpretation of in situ signatures. The results also indicate means to increase geo-effectiveness with multiple CMEs, which can be considered as another manifestation of the "perfect storm" scenario proposed by [liu14a] although the current case is not "super" in the same sense as the 2012 July 23 event.

physics.space-ph↗

arXiv2017

Multi-spacecraft Observations of the Coronal and Interplanetary Evolution of a Solar Eruption Associated with Two Active Regions

We investigate the coronal and interplanetary evolution of a coronal mass ejection (CME) launched on 2010 September 4 from a source region linking two active regions (ARs) 11101 and 11103, using extreme ultraviolet imaging, magnetogram, white-light and in situ observations from SDO, STEREO, SOHO, VEX and Wind. A potential-field source-surface model is employed to examine the configuration of the coronal magnetic field surrounding the source region. The graduated cylindrical shell model and a triangulation method are applied to determine the kinematics of the CME in the corona and interplanetary space. From the remote sensing and in situ observations we obtain some key results: (1) the CME was deflected in both the eastward and southward directions in the low corona by the magnetic pressure from the two ARs and possibly interacted with another ejection, which caused that the CME arrived at VEX that was longitudinally distant from the source region; (2) although VEX was closer to the Sun, the observed and derived CME arrival times at VEX are not earlier than those at Wind, which suggests the importance of determining both the frontal shape and propagation direction of the CME in interplanetary space; (3) the ICME was compressed in the radial direction while the longitudinal transverse size was extended.

astro-ph.SR↗